Hi

Thus my earlier set of posts about this point, and which CPLD's were the better 
ones to use for this sort of thing. 

Bob


On Mar 18, 2010, at 6:14 PM, Bruce Griffiths wrote:

> You need to select the CPLD with some care.
> Typically a low power CMOS CPLD uses an internal state machine to transfer 
> the configuration data sored in internal EEPROM (or similar) to internal CMOS 
> RAM cells that control the CPLD internal interconnections/routing.
> 
> This intialisation process occurs during power up.
> After this initialisation the EEPROM storage is powered off to reduce the 
> static power consumption.
> Some CPLDs turn the internal initialisation state machine oscillator off when 
> this process is complete, some do not.
> 
> Thus the internal intialisation state machine oscillator is a potential 
> source of unwanted asynchronous phase modulation of the divider outputs.
> 
> Bruce
> 
> Bob Camp wrote:
>> Hi
>> 
>> I think I'd just take the design over to a reasonable CPLD and be done it,
>> if you are trying to improve it's floor. Having everything on a single
>> "lump" of high speed silicon takes care of a lot of issues.
>> 
>> Bob
>> 
>> -----Original Message-----
>> From: [email protected] [mailto:[email protected]] On
>> Behalf Of David C. Partridge
>> Sent: Thursday, March 18, 2010 8:41 AM
>> To: 'Discussion of precise time and frequency measurement'
>> Subject: Re: [time-nuts] Frequency divider PCB: Current status
>> on"pre-orders", and pointers to documentation.
>> 
>> I've just had in interesting offline chat with Warren S.
>> 
>> Basic outline is that this divider is plenty good enough to use with e.g. a
>> TBolt (which was my design target), and probably also for the less than
>> totally committed time nuts.
>> 
>> To quote him:
>> 
>> "What I'm saying is does not meet the Nut cases description of a low noise
>> buffer / divider."
>> For good performance, do make absolutely certain that the power supply for
>> this board isn't fed from the same PSU as your Rb or GPSDO, or if it is,
>> make sure you take the regulated 12V from that supply and feed this board
>> through another stage of cleanup using a 5V linear regulator and decoupling
>> caps right close to the power input connector.   Linear supply rather than
>> switched (I admit that mine is switched, then post regulated).  Star
>> grounding of course.
>> 
>> But for the ultimate time nuttery level of low jitter (low pS level), the
>> following changes would NEED to be made.
>> 
>> 1) Transformer coupled input clock shaper with multi-stage filtering to
>> restrict input fed to comparator to only the intended single input frequency
>> (10MHz).
>> 
>> 2) Change the power supply to the board to be (say) 12V or 9V.  Then use
>> several 5V on board regulators to feed:
>> 
>>  a) the clock shaper
>>  b) the main '163 divider
>>  c) the 4017 divider chain
>>  d) each 74AC541 output driver and related FFs.
>> 
>> Did I get that right Warren?
>> 
>> This ensures that if e.g. the switched output is set for 1Hz, that the 5V
>> supply for the other parts of the board doesn't twitch in sympathy when the
>> '541 switches.
>> 
>> 3) Maybe use BIG earth bond terminal and feed +ve supply in on single
>> header.
>> 
>> Warren, please jump in to add further points
>> 
>> Warren is of the view that using COGs and thin film resistors isn't
>> necessary except possibly to use thin film parts for R4&  R5
>> 
>> Regards,
>> David Partridge
>> Email:[email protected]
>> 
>> -----Original Message-----
>> From: [email protected] [mailto:[email protected]] On
>> Behalf Of David C. Partridge
>> Sent: 17 March 2010 18:28
>> To: [email protected]; [email protected];
>> [email protected]; [email protected];
>> 'Discussion of precise time and frequency measurement'
>> Subject: [time-nuts] Frequency divider PCB: Current status on
>> "pre-orders",and pointers to documentation.
>> 
>> The current situation is that I have almost enough statements of intent to
>> get to the magic 50 which will allow a price of GBP14.50 per board plus
>> delivery.   For the avoidance of doubt, this is the price for a bare PCB,
>> not for a kit, and definitely not for a made up board.
>> 
>> I intend to "keep the book open" until 18:00 Zulu (UTC or GMT) on Sunday
>> 21st March, I will then count up what I have and order that many boards (and
>> maybe a few over to get a nice round number).
>> 
>> I've received numerous reqeusts for the design documentation, schematic, and
>> a bill of materials
>> 
>> They can all be downloaded from my website, but there's no way (yet) to
>> navigate to them (a round tuit problem).
>> 
>> Write up:
>> 
>> <http://www.perdrix.co.uk/FrequencyDivider/Frequency%20Divider%202.pdf>
>> 
>> Schematic:
>> 
>> <http://www.perdrix.co.uk/FrequencyDivider/Frequency%20Divider%202%20Schemat
>> ic.pdf>
>> 
>> and BOM:
>> 
>> <http://www.perdrix.co.uk/FrequencyDivider/Frequency%20Divider%202%20Bill%20
>> of%20Materials.pdf>
>> 
>> The schematic and write up have both been updated today, and the BOM is new
>> today.
>> 
>> For those who worry about SMT soldering, you don't need a reflow oven, it
>> can all be done with tweezers, a small tipped iron, fine solder wire, and
>> liquid flux (or a flux pen).  A good pair of strong reading glasses helps
>> too!   See:
>> 
>> <http://www.curiousinventor.com/guides/Surface_Mount_Soldering/101>
>> 
>> I've also had questions on part pricing:  Back in 2008, the cost to populate
>> one PCB using a MAX999, thick film resistors, and standard (X7R) chip
>> capacitors was about GBP28 including Molex headers and SMB sockets.   I
>> don't expect it to be massively different now.   I'm afraid I don't have
>> full parts kits, and the necessary up front costs to do so is more than my
>> finances allow at present.
>> 
>> FWIW, the ADCMP600 is a bit pricier than the MAX999, and is supposed to be
>> "better", though I'm not sure in what respects it is better.
>> 
>> If you want the lowest possible level of phase noise, you would follow the
>> bill of materials recommendations and use thin film resistors and C0G
>> capacitors in the clock shaper part of the circuit at the very least, but
>> this adds considerably to the cost (for example 100nF C0G 1206 capacitors
>> are about 1 pound each, while an X7R part is only a few pence).
>> 
>> Regards,
>> David Partridge
>> 
>> 
>> _______________________________________________
>> time-nuts mailing list -- [email protected] To unsubscribe, go to
>> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>> and follow the instructions there.
>> 
>> 
>> _______________________________________________
>> time-nuts mailing list -- [email protected]
>> To unsubscribe, go to
>> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>> and follow the instructions there.
>> 
>> 
>> 
>> 
>> _______________________________________________
>> time-nuts mailing list -- [email protected]
>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>> and follow the instructions there.
>> 
>>   
> 
> 
> 
> _______________________________________________
> time-nuts mailing list -- [email protected]
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
> 


_______________________________________________
time-nuts mailing list -- [email protected]
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Reply via email to